How to read an existing dataframe

[code4indo]

🐛 Describe the bug

The generated code expects a file_name.csv, as shown in the example below, How to read an existing dataframe. i'm running in jupyter notebook

Running PandasAI with openai LLM...

Code generated:

import pandas as pd
import matplotlib.pyplot as plt
df = pd.read_csv('nama_file.csv')
suspicious_df = df[df['Is Laundering'] == 1]
for currency in suspicious_df['Receiving Currency'].unique():
    currency_df = suspicious_df[suspicious_df['Receiving Currency'] == currency
        ]
    plt.scatter(currency_df['Amount Received'], currency_df['Amount Paid'],
        label=currency)
plt.xlabel('Amount Received')
plt.ylabel('Amount Paid')
plt.legend()
plt.show()

[gventuri]

@code4indo thanks for reporting. Would you mind sharing the prompt you used?

[code4indo]

@code4indo thanks for reporting. Would you mind sharing the prompt you used?

pandas_ai.run(df, prompt="Create a scatterplot showing the relationship between the amount of money received and the amount of money paid in transactions suspected of money laundering, based on the currency used")

[code4indo]

for testing purpose i using this dataset https://www.kaggle.com/datasets/ealtman2019/ibm-transactions-for-anti-money-laundering-aml

[yzaparto]

@code4indo Could you share the dataset as well if its public or a mock version?

[code4indo]

@code4indo Could you share the dataset as well if its public or a mock version?

You can download CSV data from link above

[yzaparto]

@code4indo @gventuri The code generated by the llm is wrong. It should not have the line df = pd.read_csv('nama_file.csv'). Have created a pr to try to alleviate this problem by providing a better prompt that links the question with the df already supplied.

[hunaone]

hello , i use pycharm to run the code, they run successfully, but they don't show the result,what should i do? thank u

[gventuri]

Closing, as the hallucination that overrides the df has now been fixed